MLG 032 Cartesian Similarity Metrics
Machine Learning Guide
00:00
How to Cluster Word Vectors and Document Vectors
We don't care the magnitude as much in the case of comparing word vectors or document vectors. We care more about their angular distance to each other. What i am doing in nothe is you have your journal entries. They get imbedded to vectors using u k p leb sentence transformers. And then i do the same for books from a book data base, based on their descriptions,. And then now we have two vectors, a and b. I use the cosign similarity between your entry and a book, and then i find the most similar books to your entries, and i take to top 30, and those will be your book recommendations.
Transcript
Play full episode
